CoBRA
[🏆 CHI26 Best Paper] CoBRA: Reproducible control of LLM agent behavior via classic social science experiments
- 99
- Python
- MIT
- 7 mo ago
AISmithLab
AI Smith Lab at UCSD. Toolsmiths for the age of agentic AI. Human x AI Systems × Privacy × Security x Trust.
Built from AISmithLab's public GitHub profile. Not on Take My Repo yet, so there are no posts or updates here.
[🏆 CHI26 Best Paper] CoBRA: Reproducible control of LLM agent behavior via classic social science experiments
HumanStudy-Bench: Towards AI Agent Design for Participant Simulation
Panopticon is an educational board game that teaches data science students privacy design skills through interactive gameplay.
PersonalDataHub: Gate personal data access for AI agents
Articulating privacy design by answering LLM-generated multiple-choice questions
Can we deliberately degrade an LLM’s reasoning ability in a small set of high-risk domains—such as bomb-making—while preserving its overall performance in general use?
Distribution artifacts for the ilovebenchmark desktop app. No source code lives here.
A PGS Catalog Server put together following the official guide from MIS.
A synthetic healthcare appointment scheduling benchmark for evaluating AI agents.
PEPR 2026 talk: Turning Privacy Risk Assessment into 20 Questions for Developers
We ran polygenic risk score (PRS) calculations using 1000 Genome Project's Omni Microarray genotype data on a population of 2318 individuals and show the ancestry and PRS results here.